During data transmission, to increase a data throughput, a MIMO (Multiple Input Multiple Output) system may be used, so that nt transmit antennas of a transmit end send data separately, and at the same time, a receive end uses nr receive antennas to receive and restore original data. Both nt and nr are integers greater than or equal to 2. Data is encoded and modulated in a unit of code word, and a code word includes at least one data stream. A quantity of data streams in the MIMO system is less than or equal to a smaller value between nt and nr.
In the prior art, nt transmit antennas of a network node (for example, a base station) send data separately, and nr receive antennas of corresponding UE (User Equipment, user equipment) receive the data, and then the UE performs restoration processing on the data. The UE needs to perform the restoration processing on the received data according to a sequence of receiving the data by the receive antennas. As shown in FIG. 1, after receiving the data, the UE first obtains a code word 1; then, performs MIMO equalization 1 (that is, MIMO equalization processing) and channel decoding 1 (that is, channel decoding processing) on code word 1, to obtain processed data 1 after completing the MIMO equalization 1 and the channel decoding 1; and completes processing on code word 1. Subsequently, the UE performs cancellation processing on the processed data 1 and data of a code word 2 to obtain cancellation data 1, and then performs MIMO equalization 2 and channel decoding 2 on the cancellation data 1, so as to obtain processed data 2. By analogy, processing on all code words is finally completed according to the foregoing method.
However, in a process of processing a code word by UE, a previous code word identification result is required for identifying a subsequent code word. Therefore, a code word with good channel quality needs to be preferentially used for identification, to obtain a code word identification result as accurate as possible. However, when there is an excessively great channel quality difference between streams in a code word, estimated code word quality is inaccurate or overall quality of all code words is basically similar, and as a result, a code word with good channel quality cannot be preferentially identified. Consequently, a problem easily occurs in previous code word identification, and then an error occurring in processing of a subsequent code word increases, and processing on the subsequent code word may fail.